This paper analyzes the length of enrollment in Health Maintenance Organizations (HMOs) by performing secondary data analysis. It estimates a Weibull survival function using only simple regression. For middle-class enrollees in one HMO, the hazard rate (the disenrollment rate for the remaining members of a cohort) falls sharply with the length of enrollment. For Medicaid beneficiaries allowed to remain in the fee-for-service sector, the hazard rate in another HMO is much greater but falls less sharply. Limiting the choice of Medicaid beneficiaries to HMOs would increase their length of enrollment, lowering a major barrier to HMOs' participation in Medicaid.